Normalizing the similarity score provided by gensim method most_similar_cosmul

Question

I am trying to decide whether to use the gensim methods most_similar() and most_similar_cosmul() for a project where I am trying to find a set of words that are similar to a